Results

Allele frequencies

The frequency of the 249 bp allele at M6S161 varied considerably among family cohorts (Table Ia, complete allele frequency distributions for all cohorts and confidence intervals for the allele frequencies determined for the family cohorts are accessible at URL http://www.psoriasis.umich.edu/cluster17/.) Even though we have not randomly sampled the populations of these countries, the observed lack of association of the 249 bp allele with psoriasis (see below) indicates that the observed frequencies are probably a reasonable estimate of the actual frequencies in the general population. It has been shown by others that the Cw8-B65 haplotype of cluster 17 is found more frequently along the Mediterranean than other parts of Europe (^{Imanishi et al, 1992}). This finding is borne out by our data, where frequencies less than 1.4% are seen in Sweden, Finland, Iceland, Germany, and England, and frequencies greater than 2.1% are seen in Italy, France, and the United States (the Ann Arbor/Kiel sample is largely from the United States, which is home to many individuals of Mediterranean origin).

Table I - Summary statistics for pedigrees and case–control samples used in collaborative study of risk for psoriasis of cluster 17 haplotype (M6S161 allele 249–HLA-Cw8–HLA-B65).

Full table

The two case–control samples are summarized in Table Ib. Note that the frequency of allele 249 in the Ann Arbor and Reykjavik case–control samples is very similar to its frequency among founders in the pedigree samples for these same two countries (1.2%vs 1.4% for Reykjavik; 2.1%vs 2.1 and 2.6% for Ann Arbor).

Power analysis

Power to detect association of cluster 17 and psoriasis depends in part upon the genotype relative risk (GRR) for developing psoriasis when carrying one or two copies of the cluster 17 haplotype. If cluster 17 is truly associated with psoriasis, then recombinant haplotype analysis of the original Ann Arbor/Kiel cohort (^{Nair et al, 2000}) strongly suggests that the cluster 17 haplotype and all Cw6 haplotypes carry the same PSORS1 risk allele within a segment of DNA that is inherited identical by descent from a common ancestor. Accordingly, an estimate of the GRR of a putative cluster 17 PSORS1 risk allele can be provided by estimating the GRR of Cw6 in our sample. We base the GRR estimate on Cw6 haplotypes only rather than on cluster 17 by itself or in combination with Cw6, because multiple studies have provided incontrovertible evidence that Cw6 haplotypes carry a PSORS1 risk allele, whereas the evidence for cluster 17 carrying PSORS1 is limited to one result of marginal significance.

Estimates of GRR for Cw6 in the original Ann Arbor/Kiel cohort (^{Nair et al, 2000}) are 10.4 for homozygotes and 5.2 for heterozygotes, indicating that the relationship of psoriasis and PSORS1 approximately follows an additive mode of inheritance. These risks, however, are expressed relative to the risk of non-Cw6 carriers, whereas the power calculations for cluster 17 assume risk is expressed to non-cluster 17 carriers, many of which will carry at least one Cw6 haplotype. Including Cw6 carriers among the "non-risk" genotype category nearly doubles the penetrance of "non-risk" genotypes and thus halves the expected GRR for the putative cluster 17 PSORS1 allele. Accordingly, a reasonable estimate of GRR for cluster 17 as a PSORS1 risk haplotype would be 5 for GRR2 and 3 for GRR1.

These GRR values were used to perform power calculations as described in Subjects and Methods. The new cohort has 791 pedigrees with at least one triad, and the frequency of cluster 17 is assumed to be 0.015, the observed frequency of M6S161 allele 249 in the new pedigrees. For an additive model with a GRR2 of 5 (GRR1=3), which is reflective of the actual data, and a type I error rate of 0.05, power of the two-sided TDT is 99.8%. Importantly, 80% power is still achieved even if the relative risk of cluster 17 is much lower than that seen for Cw6 haplotypes. Thus, 80% power is achieved for GRR2=3.03 under an additive model (GRR1=2.01). Moreover, the true power of our pedigree disequilibrium test (PDT) analysis of full pedigrees is no doubt substantially higher than our estimates, which consider only TDT analysis of independent triads. Results are broadly similar for dominant and multiplicative models; however, power is much lower for recessive models (6% for a type I error rate of 0.05, GRR2=5, GRR1=1). Complete power curves are available at URL http://www.psoriasis.umich.edu/cluster17/.

Our original pedigrees showed marginal evidence for association (TDT p value=0.089), and our power calculations predict that we should see stronger evidence for association (i.e., a smaller p-value) with a probability of >99.9% if allele 249 was associated with psoriasis at a GRR2 of 5, assuming an additive or dominant mode of gene action. Under a multiplicative model, this probability is still >95%.

Power was also determined for TDT analysis of the new and old samples combined. For the pooled pedigree sample, there are 1167 independent triads, and the observed frequency of M6S161 allele 249 is 0.0167. Power of the two-sided TDT to detect association of psoriasis and cluster 17 for a GRR2 of 5 and type I error rate of 0.05 is essentially 100% under the additive or dominant models, 99% under a multiplicative model, and 6% under a recessive model. 80% power is achieved if GRR2=2.52 (GRR1=1.76) under the additive model. As expected, the power of the pooled sample to detect association is substantially greater than the new sample alone. Thus, both the new and combined samples appear adequately powered, and we would expect to replicate our original association if M6S161 allele 249 were truly associated with an increased risk of psoriasis.

Family-based analyses

Results of the TDT analysis are shown in Table II. The original cohort described by^{Nair et al (2000)} suggests association of psoriasis with allele 249: a 65.7% transmission ratio, and p-values of 0.089 (exact) and 0.063 (asymptotic) (Table IIa). The number of transmission events is higher here (23:12) than in the original publication (21:10) because the 249 allele occurs on a few haplotypes that were directly adjacent to but not actually within cluster 17 in the clustering dendrogram. Inclusion of these haplotypes is justified because they are completely homologous to the cluster 17 consensus for those markers that lie within or near the PSORS1 candidate region. The new pedigrees, however, show no evidence for association of cluster 17 with psoriasis (T:NT=21:21, p=1.0). Combining the new pedigrees with the original cohort also yields no significant association (43:32, p=0.25).

Table II - Results of biallelic TDT test (median of 999 runs).

Full table

Results of the PDT analysis for allele 249 are given in Table III. The original cohort shows some evidence for association of cluster 17 with psoriasis, as expected. When analysis is restricted to triads and dyads, the asymptotic p-value gives somewhat stronger evidence for marker-trait association than was determined by the TDT (0.041 vs 0.063), even though the percent transmission is considerably lower (60.9%vs 65.7%). Analysis of discordant sib pairs (DSP) in the original cohort also shows evidence for association (= 0.247, see Subjects and Methods for definition of ). When triads, dyads, and DSP are combined, the original cohort shows considerably more significant association of cluster 17 and psoriasis by the PDT (= 0.219, p=0.027) than by the TDT. PDT analysis of the new pedigrees, however, shows no evidence for association of cluster 17 and psoriasis, whether the analysis uses triads and dyads (= 0.025, %T = 51.3, p=0.80), DSP (=0.018, p=0.98), or both (=0.019, p=0.69). The combined cohorts also show no significant association of psoriasis and allele 249. PDT analysis of triads, dyads, and DSP, however, yields a result that approaches p=0.05 (=0.108, p=0.083). Although the original cohort comprises only 29.9% of the 1618 pedigrees amenable to PDT analysis, it makes up 42.4% of the 92 pedigrees in the total cohort that are actually informative for allele 249 on account of the higher frequency of the 249 bp allele in the original cohort versus the new pedigrees (2.1%vs 1.5%).

Table III - Results of PDT analysis for cluster 17 (M6S161 allele 249).

Full table

In contrast to the negative results obtained for allele 249, TDT analysis of the 248 bp allele at M6S161, a common allele that is found on all HLA-Cw6-positive risk haplotypes as well as several non-risk haplotypes, shows an excess transmission for all cohorts, ranging from 54.4% to 66.7% (Table IIb). The association is moderately significant for the original Ann Arbor/Kiel cohort (197:151, p=0.016) and highly significant for the new pedigrees combined (431:294, p=4.1 10^-7) and for the total sample (630:447, p=2.7 10^-8). The association of allele 248 with psoriasis reflects the fact that this allele is carried on all HLA-Cw6-positive haplotypes. The association of psoriasis with allele 248 (58.5% transmission, Table IIb) is lower than that seen for HLA-Cw6-positive haplotypes (73.6% transmission (^{Nair et al, 2000})) because allele 248, by far the most frequent allele of M6S161, is also found on many common non-risk haplotypes.

Results for PDT analysis of allele 248 are shown in Table IV. As was observed using the TDT (Table IIb), highly significant evidence of association of psoriasis and allele 248 is seen for the original, new, and combined cohorts. PDT analysis of triads and dyads gives more striking evidence for association than TDT analysis (p=0.0070 vs 0.014 for the original cohort, 4.5 10^-8vs 3.6 10^-7 for the new cohort, and 1.4 10^-9vs 2.5 10^-8 for the total cohort), although the percent transmission values are very similar (57.1%vs 56.6% for original, 59.4%vs 59.5% for new, and 58.6%vs 58.5% for total cohort). Inclusion of DSP yields even greater significant evidence for association of psoriasis and the 248 bp allele. Although the p-values for DSP analysis of allele 248 are less impressive than for triads and dyads because of the smaller numbers of informative pedigrees, the levels of disequilibrium for triads and dyads versus DSP are comparable (=0.202 vs 0.141 for the original cohort, 0.133 vs 0.187 for the new cohort, and 0.158 vs 0.172 for the total cohort). Hence, concerns that PDT analysis of DSP would be adversely affected by the relatively low penetrance of PSORS1 (i.e., that unaffected siblings of psoriatics may often be carriers for PSORS1 haplotypes) are apparently unwarranted.

Table IV - Results of PDT analysis for M6S161 allele 248.

Full table

Case–control analysis

Table Va gives results for the case–control analysis of allele 249 (cluster 17). Applying the trend test to the three genotype categories found no significant evidence for association of disease and allele 249 within the Ann Arbor cohort (odds ratio (OR)=0.31, p=0.12) or the Reykjavik cohort (OR=1.38, p=0.45). For these data, the "allele" and "serological" tests (^{Sasieni, 1997}) would have given exactly the same p-values as the "genotype" test because there are no homozygotes for the 249 allele. Although the OR differ for the two cohorts, there is no significant evidence for heterogeneity in the association of genotype and disease across cohorts (p=0.093). The Mantel extension of the linear trend test found no significant association of allele 249 and psoriasis across both cohorts (p=0.64). Thus the case–control analysis confirms the results of the family-based tests.

Table V - Results of case–control analysis.

Full table

Table Vb lists results for case–control analysis of allele 248, which is found on all Cw6-positive haplotypes as well as many common non-risk haplotypes. The Ann Arbor cohort shows significant evidence for association of allele 248 with psoriasis (p=0.015), whereas the Reykjavik cohort fails to reach the threshold for significance (p=0.090). Linear association appears to be homogeneous across cohorts (p=0.31), so it is appropriate to test for conditional independence. As expected, the Mantel extension of the trend test across both cohorts provides more power than the individual cohort tests, finding strong evidence for association of allele 248 genotype and psoriasis (p=0.0053). Thus, once again, analysis of case–control data confirms the results of the pedigree analysis.

Note that the OR for homozygotes is greater than that for heterozygotes in both cohorts (2.74 vs 2.05 in Ann Arbor cohort; 1.93 vs 1.75 in Reykjavik cohort), and that the difference in risk of carrying two versus one copy of the 248 bp allele most closely resembles an additive model. Similar results have been reported recently by some of us (^{Gudjonsson et al, 2003}).

Discussion

A critical task in the identification of the PSORS1 gene is to identify the shortest genetic interval that contains it. This task is challenged by the substantial linkage disequilibrium that characterizes the human MHC (^{Walsh et al, 2003}). This collaboration was undertaken because cluster 17 (HLA-Cw8-B65) was suggested as associated in a previous study, albeit with a marginally significant p-value of 0.046 (^{Nair et al, 2000}). This cluster displayed great similarity to other risk-associated haplotypes in the proximal MHC Class I region, yet it lacked HLA-Cw6 (^{Nair et al, 2000}). When combined with subsequent sequence information (see Introduction), it became evident that the length of the risk interval depended critically on a more definitive assessment of the risk associated with cluster 17. As shown in Figure 1, if cluster 17 is not a genuine risk haplotype, the candidate region for PSORS1 expands from a 140 kb region encompassing four genes to a 300 kb region encompassing eight genes.

Comparison of haplotype clusters revealed that the 249 bp allele at M6S161 was specific for cluster 17. Indeed, marker M6S161 and the flanking genes OTF3, TCF19, HCR, SPR1, SEEK1, CDSN, and STG reside within one of the seven strongest regions of linkage disequilibrium in the MHC (^{Walsh et al, 2003}). This finding greatly facilitated collaboration, because it was possible for each participating center to type only a single marker as a surrogate for cluster 17. Although different typing methodologies were used by different groups, the provision of DNA from a single reference individual (a 248 bp/249 bp heterozygote at M6S161) allowed for accurate calibration of allele sizes.

The replication set formed by this collaboration provided no evidence for association between psoriasis and cluster 17 in the TDT (Table IIa), the PDT (Table III), or in the trend test for case–control genotypes (Table V). Even when combined with the original cohort that yielded a marginally significant result (^{Nair et al, 2000}), the data also fail to identify a significant association between cluster 17 and psoriasis by either the TDT or the PDT (Table IIa, Table IV, and Table V). These data are in accord with (but not fully independent of) a study of over 1000 Icelandic patients with psoriasis, which failed to identify an association between psoriasis and HLA-Cw8 (^{Gudjonsson et al, 2003}).

We have given careful consideration to the power of our sample to detect a significant association if one truly existed. Our power calculations demonstrate excellent power to detect association, under all reasonable genetic models, for realistic values of GRR. Although our data set lacked power under a recessive model, GRR values derived from our own data and association studies performed on over 1000 Icelandic psoriatics (^{Gudjonsson et al, 2003}) strongly suggest that PSORS1 does not act in a recessive fashion. Our power calculations are conservative, in that they include only a subset of the individuals and genetic relationships actually used by our association tests.

Taken together, the findings presented here demonstrate that cluster 17 is highly unlikely to encode a PSORS1 risk allele. Although this result extends the length of the risk haplotype from 140 to 300 kb (Figure 1), it also provides a much larger number of candidate genes on that haplotype (eight) than did the previous interval (four). Of these, HLA-Cw6, the "WWCC" allele at HCR, and "allele 5" at CDSN remain particularly plausible, whereas the evidence in favor of the remaining genes is less robust (^{Asumalahti et al, 2002;Capon et al, 2002}). The paucity of recombination between these alleles has made it extremely difficult to identify enough informative recombinant individuals to distinguish the risk associated with each of them. Although transracial mapping of a Gujarati Indian population has suggested a co-equal role for HLA-Cw6 and CDSN, but not for HCR (^{Capon et al, 2003}), other studies of Japanese, Thai, Chinese, and Spanish psoriatics have failed to confirm the importance of CDSN allele 5 independently of HLA-Cw6 (^{Gonzalez et al, 2000;Hui et al, 2002;Chang et al, 2003;Romphruk et al, 2003}). Two additional studies of HCR have suggested that the WWCC allele at HCR is unlikely to play a genetically causal role in psoriasis, even though it is highly associated with the disease (^{Chia et al, 2001;O'Brien et al, 2001}). HCR, however, was recently functionally assayed in transgenic mouse models, and the risk allele WWCC was suggested to induce allele-specific effects on the expression of some genes relevant for psoriasis, even though the animals remained healthy (^{Elomaa et al, 2004}). Clearly, additional very large studies involving multiple ethnic/racial groups will be required to distinguish between these three attractive candidate genes. This study demonstrates that such collaborations are feasible and can be productive. An ongoing study involving ourselves and others will analyze thousands of psoriatics, family members, and controls, making use of the emerging resources of the^{HapMap project (2003)} in an effort to address this important problem. The use of markers carrying alleles that are specific for particular haplotype blocks, or, as in this analysis, for particular combinations of haplotype blocks, will undoubtedly lead to substantial savings of effort, time, and cost.

Subjects and Methods

Subjects

In addition to the samples supplied by each group of European collaborators, the Ann Arbor/Kiel group has continued to collect clinical material (families, single affected individuals with parents, and cases and controls) since the time of our original publication (^{Nair et al, 2000}). As shown in Table Ia, these sources combined to provide an additional 1275 pedigrees for association testing. The number of pedigrees is shown for each cohort, along with the subset of this number that contains at least one triad (affected child with two typed parents), at least one dyad (affected child with one typed parent) but no triads, or at least one DSP. By any measure, the new sample contains more than twice the number of pedigrees in the original cohort of^{Nair et al (2000)}. Combined, the two samples yield a total of 1544 pedigrees for TDT analysis and 1618 pedigrees for PDT analysis.

The Ann Arbor and Reykjavik groups were also able to provide case–control samples (Table Ib). Together, these two samples provide 300 cases and 913 controls, allowing a second independent test for association of cluster 17 and psoriasis.

Enrollment of subjects and genotyping was carried out under protocols approved by the medical ethical committees of each participating institution. Written, informed consent was obtained from all subjects. This study was conducted according to the Declaration of Helsinki Principles at all participating institutions.

Marker typing

All members of the pedigree and case–control samples were typed for microsatellite marker M6S161, which is one of the markers used to create the 34-marker haplotypes described by^{Nair et al (2000)}. This marker carries an allele of size 249 bp that is highly specific to the cluster 17 haplotype. All of the other risk clusters (19, 21–23, 25) carry the 248 bp allele at M6S161, which is the most common allele at this marker and is also found on many non-risk haplotypes. The M6S161 amplimer produces several different one-nucleotide variations in allele size, because it includes two different length polymorphisms—a TC dinucleotide repeat and a 1 bp C/-indel (see foonote 2). Each group performed its own genotyping, utilizing ³²P-labeled or fluorescent oligonucleotide primers by standard methods (^{Nair et al, 1995;Veal et al, 2001}). DNA from a reference individual known to be heterozygous for the M6S161 248 and 249 bp alleles was genotyped in parallel by all groups, allowing calibration of results and comparison of allele sizes across centers. Each genotyping center employed its own established procedures to minimize genotyping errors. Whenever possible, genotyping accuracy was further assessed by checking for Mendelian inheritance errors.

Family-based association analysis

The pedigrees were analyzed for M6S161 alleles 248 and 249 by two different family-based association tests—the TDT (^{Spielman et al, 1993}) and the PDT (^{Martin et al, 2000;Martin et al, 2001}). We utilized the "PDT-avg" test described by^{Martin et al (2001)}, which gives equal weighting to all families, rather than the "PDT-sum" test, which gives greater weight to larger pedigrees. Both the TDT and PDT were extended to include dyads when triads were not available.

The value is a scaled measure of linkage disequilibrium between the allele being tested by the PDT and the disease phenotype. It is an alternative to reporting the percentage of transmitted alleles in the TDT. has a range of [-1, 1] and is equal to 0 in the absence of linkage disequilibrium. The original description of the PDT (^{Martin et al, 2000}) provided no standardized measure of the level of linkage disequilibrium measured by the PDT statistic. Martin et al do define a random variable D, where D_i summarizes the amount of linkage disequilibrium for all possible triads and DSP for the ith pedigree in a sample of N-independent pedigrees. The range of permissible values of D, however, depends upon the structure of the pedigree and upon the parental genotypes in the triads, dyads, and DSP (we have extended the analysis to dyads, as mentioned above). In order to standardize the D_i for each pedigree, we divided it by the maximum possible D value (conditional on known parental genotypes) for that pedigree. This restricts the range of D to the [-1,1] interval. Mean standardized D values were then computed over all analyzed pedigrees.

Many of the pedigrees in the sample are large nuclear families or extended families that contain more than one triad and/or dyad. Because the TDT is a valid test of association only when the triads and dyads analyzed are genetically independent, analysis was restricted to a single triad or dyad randomly selected from each pedigree. Since results vary depending upon the particular random selection, the analysis was repeated 999 times with different random number seeds, and the median result reported. To avoid bias, we restricted the analysis of dyads to instances where the parent has a heterozygous genotype different from that of the affected child (^{Curtis and Sham, 1995}). Although TDT and PDT were both run as biallelic tests, determination of alleles transmitted and non-transmitted from heterozygous parents was based on the full allele diversity of the marker. This permits use of at least some of the dyads, which would never qualify for unbiased analysis if the marker alleles were first downcoded to a biallelic system (^{Curtis and Sham, 1995}).

The biallelic PDT-avg test statistic is asymptotically distributed as a ² with one degree of freedom. Unlike the biallelic TDT test statistic, it is not possible to compute an exact binomial p-value for the PDT. Hence, for comparison purposes, asymptotic p-values will be used whenever TDT and PDT results are compared.

Cohorts were analyzed individually and in combination (original cohort, new cohorts combined, and all cohorts combined). We considered it permissible to directly analyze the combined data with family-based tests, because the validity of such tests is not affected by population stratification or admixture for the locus being tested (^{Spielman et al, 1993;Martin et al, 2000}).

Case–control association analysis

As recommended by^{Sasieni (1997)}, we have analyzed our case–control data as genotypes, in conjunction with the Cochran–Armitage test for linear trend. The assumptions of the linear trend test fit the data most optimally under a multiplicative model, but the trend test is appropriate as long as the risk for individuals with two copies of the disease-associated allele is not intermediate between those who carry one or zero copies. This is certainly the case for PSORS1, which seems to follow an additive model (see below). Unlike family-based association tests such as the TDT and PDT, case–control methods are sensitive to population stratification, so we did not directly combine data for the Ann Arbor and Reykjavik cohorts. Mantel's extension of the Mantel–Haenszel test to allow for ordinal variables (^{Mantel, 1963}) was used to test for the conditional independence of psoriasis and M6S161 genotype while controlling for cohort. Genotype categories were assigned scores for the number of test alleles (0, 1, 2), which is the implicit scoring system of the Cochran–Armitage trend test. The Mantel test works best when linear association is similar across strata (cohorts); for this reason, a likelihood ratio test for homogeneous linear association was constructed by comparing the logistic regression models that include and exclude an interaction term between genotype and cohort.

Power analysis

Our sample consists of a variety of family structures, for which analytical power calculations are unavailable. To simplify the power analysis, we determined the power of the TDT to detect association when considering only genetically independent triads of the new and pooled cohorts, using the first approximation method of^{Knapp (1999)}. This provides a conservative estimate of the actual power of our pedigree sample, since it discards all dyads (used by the TDT and PDT), additional triads in the family (used by the PDT), and DSP (used by the PDT). This method also allowed us to compute power for a reasonable range of alternative hypotheses, basing the alternatives on what we know about the likely GRR of PSORS1, the observed frequencies of the 249 bp allele in our cohorts, and reasonable genetic models (i.e., recessive, dominant, multiplicative, and additive).

We estimated the penetrance, or the probability of disease for carriers of a particular genotype, using estimates of genotype frequencies from our previous study of PSOR1 (^{Nair et al, 2000}), and from 189 US organ transplant donor candidates and 124 German blood donors (^{Jenisch et al, 1998}), each weighted by the relative contribution of each country to our cohort. We assumed a population prevalence of 2% for psoriasis. We then used these penetrance estimates to define GRR, the penetrance of a designated risk genotype divided by the penetrance for a genotype that carries zero copies of the allele of interest, which is the key parameter for the formulas of^{Knapp (1999)}.

Notes

While this manuscript was under review, a study of Sardinian psoriasis by Orru (Am J Hum Genet 76:164–171) also found no evidence for association between psoriasis and Cluster 17 (designated haplotype H in their study).

² Nair R, Stuart P, Voorhees J, Weichenthal M, Jenisch S, Christophers E, Elder JT: Manuscript, in preparation.

References

1. Asumalahti, K, Veal, C, Laitinen, T, et al: Coding haplotype analysis supports HCR as the putative susceptibility gene for psoriasis at the MHC PSORS1 locus. Hum Mol Genet 2002 11:589–597, 10.1093/hmg/11.5.589 | Article | PubMed | ISI | ChemPort |
2. Capon, F, Munro, M, Barker, J, Trembath, R: Searching for the major histocompatibility complex psoriasis susceptibility gene. J Invest Dermatol 2002 118:745–751,  | Article | PubMed | ISI | ChemPort |
3. Capon, F, Toal, IK, Evans, JC, et al: Haplotype analysis of distantly related populations implicates corneodesmosin in psoriasis susceptibility. J Med Genet 2003 40:447–452,  | Article | PubMed | ISI | ChemPort |
4. Chang, YT, Tsai, SF, Lee, DD, et al: A study of candidate genes for psoriasis near HLA-C in Chinese patients with psoriasis. Br J Dermatol 2003 148:418–423,  | Article | PubMed | ISI | ChemPort |
5. Chia, NV, Stuart, P, Nair, RP, et al: Variations in the HCR (Pg8) gene are unlikely to be causal for familial psoriasis. J Invest Dermatol 2001 116:823–824,  | Article | PubMed | ISI | ChemPort |
6. Consortium: The International HapMap Project. Nature 2003 426:789–796, 10.1038/nature02168 | Article |
7. Curtis, D, Sham, PC: A note on the application of the transmission disequilibrium test when a parent is missing [letter]. Am J Hum Genet 1995 56:811–812,  | PubMed | ISI | ChemPort |
8. Elder, JT, Nair, RP, Guo, SW, Henseler, T, Christophers, E, Voorhees, JJ: The genetics of psoriasis. Arch Dermatol 1994 130:216–224, 10.1001/archderm.130.2.216 | Article | PubMed | ISI | ChemPort |
9. Ellegren, H: Microsatellite mutations in the germline: Implications for evolutionary inference. Trends Genet 2000 16:551–558, 10.1016/S0168-9525(00)02139-9 | Article | PubMed | ISI | ChemPort |
10. Elomaa, O, Majuri, I, Suomela, S, et al: Transgenic mouse models support HCR as an effector gene in the PSORS1 locus. Hum Mol Genet 2004 13:1551–1561, 10.1093/hmg/ddh178 | Article | PubMed | ISI | ChemPort |
11. Gonzalez, S, Martinez-Borra, J, Del Rio, JS, Santos-Juanes, J, Lopez-Vazquez, A, Blanco-Gelaz, M, Lopez-Larrea, C: The OTF3 gene polymorphism confers susceptibility to psoriasis independent of the association of HLA-Cw*0602. J Invest Dermatol 2000 115:824–828, 10.1046/j.1523-1747.2000.00133.x | Article | PubMed | ISI | ChemPort |
12. Gudjonsson, JE, Karason, A, Antonsdottir, A, et al: Psoriasis patients who are homozygous for the HLA-Cw*0602 allele have a 2.5-fold increased risk of developing psoriasis compared with Cw6 heterozygotes. Br J Dermatol 2003 148:233–235, 10.1046/j.1365-2133.2003.05115.x | Article | PubMed | ISI | ChemPort |
13. Hui, J, Oka, A, Tamiya, G, et al: Corneodesmosin DNA polymorphisms in MHC haplotypes and Japanese patients with psoriasis. Tissue Antigens 2002 60:77–83, 10.1034/j.1399-0039.2002.600110.x | Article | PubMed | ISI | ChemPort |
14. Imanishi, T, Akaza, T, Kimura, A, Tokunaga, K, Gojobori, T: Allele and haplotype frequencies for HLA and complement loci in various ethnic groups. In: Tsuji K, Aizawa M, Sasazuki T (eds). HLA 1991. Proceedings of the Eleventh International Histocompatibility Workshop and Conference. 1992; Oxford, New York, Tokyo: Oxford Science Publications, p 1065–1220,
15. Jenisch, S, Henseler, T, Nair, RP, et al: Linkage analysis of HLA markers in familial psoriasis: Strong disequilibrium effects provide evidence for a major determinant in the HLA-B/-C region. Am J Hum Genet 1998 63:191–199, 10.1086/301899 | Article | PubMed | ISI | ChemPort |
16. Knapp, M: A note on power approximations for the transmission/disequilibrium test. Am J Hum Genet 1999 64:1177–1185,  | PubMed | ISI | ChemPort |
17. Mantel, N: Chi-square tests with one degree of freedom: Extensions of the Mantel–Haenszel procedure. J Am Stat Assoc 1963 58:690–700,  | ISI |
18. Martin, ER, Bass, MP, Kaplan, NL: Correcting for a potential bias in the pedigree disequilibrium test. Am J Hum Genet 2001 68:1065–1067, 10.1086/319525 | Article | PubMed | ISI | ChemPort |
19. Martin, ER, Monks, SA, Warren, LL, Kaplan, NL: A test for linkage and association in general pedigrees: The pedigree disequilibrium test. Am J Hum Genet 2000 67:146–154, 10.1086/302957 | Article | PubMed | ISI | ChemPort |
20. Nair, R, Guo, S, Jenisch, S, et al: Scanning chromosome 17 for psoriasis susceptibility: Lack of evidence for a distal 17q locus. Hum Hered 1995 45:219–230,  | PubMed | ISI | ChemPort |
21. Nair, RP, Stuart, P, Henseler, T, et al: Localization of psoriasis-susceptibility locus PSORS1 to a 60-kb interval telomeric to HLA-C. Am J Hum Genet 2000 66:1833–1844, 10.1086/302932 | Article | PubMed | ISI | ChemPort |
22. O'Brien, KP, Holm, SJ, Nilsson, S, et al: The HCR gene on 6p21 is unlikely to be a psoriasis susceptibility gene. J Invest Dermatol 2001 116:750–754, 10.1046/j.0022-202x.2001.01323.x | Article | PubMed | ChemPort |
23. Romphruk, AV, Oka, A, Romphruk, A, et al: Corneodesmosin gene: No evidence for PSORS 1 gene in North-eastern Thai psoriasis patients. Tissue Antigens 2003 62:217–224, 10.1034/j.1399-0039.2003.00056.x | Article | PubMed | ISI | ChemPort |
24. Sasieni, PD: From genotypes to genes: Doubling the sample size. Biometrics 1997 53:1253–1261,  | PubMed | ISI | ChemPort |
25. Spielman, RS, McGinnis, RE, Ewens, WJ: Transmission test for linkage disequilibrium: The insulin gene region and insulin-dependent diabetes mellitus (IDDM). Am J Hum Genet 1993 52:506–516,  | PubMed | ISI | ChemPort |
26. Trembath, RC, Clough, RL, Rosbotham, JL, et al: Identification of a major susceptibility locus on chromosome 6p and evidence for further disease loci revealed by a two stage genome-wide search in psoriasis. Hum Mol Genet 1997 6:813–820, 10.1093/hmg/6.5.813 | Article | PubMed | ISI | ChemPort |
27. Veal, CD, Clough, RL, Barber, RC, et al: Identification of a novel psoriasis susceptibility locus at 1p and evidence of epistasis between PSORS1 and candidate loci. J Med Genet 2001 38:7–13, 10.1136/jmg.38.1.7 | Article | PubMed | ISI | ChemPort |
28. Walsh, EC, Mather, KA, Schaffner, SF, et al: An integrated haplotype map of the human major histocompatibility complex. Am J Hum Genet 2003 73:580–590, 10.1086/378101 | Article | PubMed | ISI | ChemPort |
29. Zhang, XJ, He, PP, Wang, ZX, et al: Evidence for a major psoriasis susceptibility locus at 6p21(PSORS1) and a novel candidate region at 4q31 by genome-wide scan in Chinese Hans. J Invest Dermatol 2002 119:1361–1366, 10.1046/j.1523-1747.2002.19612.x | Article | PubMed | ISI | ChemPort |

Acknowledgments

The authors thank all the psoriasis patients, family members, and controls who have agreed to participate in the genetic analysis of psoriasis. This work was supported, in part, by an award (R01 AR 042742) from the National Institutes of Health (J. T. E., R. N., P. S., and J. J. V.) and grants from the Italian Ministry of Health (G. N.). J. T. E. is supported by the Ann Arbor Veterans Affairs Hospital.